CSLAM and GPS Based Navigation for Multi-UAV Cooperative Transportation System

Abstract

Multi-UAV cooperative technology is developing vigorously due to its convenience and expandability. In the future, more and more military or civilian transport tasks will be accomplished by multi-UAV. At present, the biggest problem of long-distance transportation is that GPS signal is not available throughout the whole journal. In this paper, we propose a complete navigation scheme for the multi-UAV cooperative transportation system in the two cases with or without GPS signal, which is designed based on visual simultaneous localization and mapping (SLAM) and GPS. If the multi-UAV cooperative system is in the GPS-denied environment, the system will make real-time positioning and mapping through collaborative SLAM (CSLAM). In this case, the relative position between the UAVs, and their absolute positions as well are available under the proposed scheme. On the contrary, if the multi-UAV cooperative transportation system can detect the GPS signal, GPS will provide a high accuracy correction of the CSLAM, and finally output more precise information compared to CSLAM only. Moreover, the navigation scheme can switch automatically according to the presence or absence of GPS signal in the process of task without human intervention. Different data sets are used for experiments, including multi-UAV CSLAM, and the fusion of CSLAM and GPS. The experiment results show the efficiency and accuracy of the proposed navigation scheme.